Stock and Marketplace Integration for Ecommerce

If the ecommerce site and a marketplace sell the same product, the stock has to be right in one place. When the last size shows as available on both, two customers buy it and one meets a cancellation. Integration is not “open two panels side by side”. It is a definition of where the order, the stock and the price are born, and how they are written to the other side.

This article does not walk key by key like a software manual. It orders the questions operations should ask. The cost of a bad integration is not technical. It is cancellations, penalty points and an angry customer. If an ecommerce website leaves these questions for phase two, the first campaign day is spent counting stock by hand.

One source of truth

One system should own the stock. That might be the warehouse software, the shop panel or the marketplace. If all three are “the real one”, the numbers argue. For most businesses the source is where the physical item actually leaves. The site and the marketplace read from it, and write back when a sale happens.

The choice also depends on which screen the team already uses. If warehouse staff never open the marketplace panel, declaring that panel the source stays on paper. In discovery we ask who enters stock, on which screen, at what time of day. If there is no answer, we design the habit before the integration.

How often does it update?

Instant updates are the ideal, and not every account can have them. Some marketplaces pull stock on an interval. Two sales can collide inside that gap. You may not be able to bring collisions to zero. You can shrink them by keeping a safety margin on critical stock and by alerting the same day when two orders clash. A promise of “no errors ever” does not fit this work. Visible errors do.

Where does the order land?

Marketplace orders and site orders should gather on one table. An order left on a separate list is the one forgotten at dispatch. The pick list should be single: product, quantity, address, shipping code. Where the invoice is cut should be written at the start too. Invoicing in two places turns accounting into a second integration problem.

• Product match: the marketplace barcode and the site SKU share one map. A match pasted by hand breaks on the first new product.
• Price: which channel a campaign price belongs to, and until when. The same discount does not have to be everywhere.
• Shipping: the code is written back to the panel where the order was born. The customer looks there, not at you, for “where is my parcel”.
• Cancel and return: if stock is not written back, the product stays ghost-sold.

Custom prices, dealers or several warehouses may outgrow a ready connector. Then the rule needs to live in code. That split also shows up in Laravel or WordPress. On the WordPress and PHP software side we do not shrug integration off as “we’ll add a plugin”. The rule first, then the tool.

No error log, no integration

When the connection drops, the system can keep showing yesterday’s stock in silence. That is the expensive kind of failure. You think it works. The customer buys a product you do not have. Each sync needs a time, the code of the product that failed, and a record that it was retried. Operations should not start the morning without looking at yesterday’s bad rows.

Test with a live product and a small stock. Do not bind a hundred-unit SKU on day one. One product, one order, one cancellation, one stock return. Do not open a second channel until those four moves pass. Putting that rehearsal on the calendar is part of what to check when building an ecommerce site.

The human side

Integration does not remove the warehouse. If the barcode is wrong, the software writes the right number onto the wrong product. If the name differs between marketplace and site, the customer disputes the parcel. Someone owns the matching table. When that person is away, the backup is named. If the panel password lives only in the agency’s pocket, you cannot see your own stock. Accounts are opened in your name. Access stays with you.

When Yuz Medya hands an integration over, a short note says which field comes from where. “It works” should not mean you cannot add a product three months later. Training is part of the connection. Adding a new marketplace is discussed like a new project: commission, the shipping deal, category matching and the error log are written again. Opening a channel by copy-paste carries the same mistakes into a second window.

Frequently asked questions

Site first, or marketplace first?

Whichever place sales already flow through gets straightened first. The stock source binds there. The other channel reads it. Opening both in the same week, with no source, produces two half operations. The order follows the screen the team actually uses, not the revenue daydream.

Does stock have to match every second?

That is the ideal, and it cannot be promised on every connection. Know the delay, leave a margin on critical items, and see colliding orders the same day. An invisible delay does more harm than a short, visible one.

If the integration breaks, should sales stop?

Automatic writes should stop. Sales do not always have to. Keeping on publishing bad stock is more expensive than pausing to check by hand. Write the threshold in advance: which error silences the sync, who is told, how a manual list is produced. Inventing the rule during the crisis is how the wrong product gets shipped.